%0 Journal Article
%9 ACL : Articles dans des revues avec comité de lecture répertoriées par l'AERES
%A Prend, C. J.
%A Keerthi, M. G.
%A Levy, Marina
%A Aumont, Olivier
%A Gille, S. T.
%A Talley, L. D.
%T Sub-seasonal forcing drives year-to-year variations of Southern Ocean primary productivity
%D 2022
%L fdi:010085334
%G ENG
%J Global Biogeochemical Cycles
%@ 0886-6236
%K OCEAN AUSTRAL
%M ISI:000819245500001
%N 7
%P e2022GB007329 [15 ]
%R 10.1029/2022gb007329
%U https://www.documentation.ird.fr/hor/fdi:010085334
%> https://horizon.documentation.ird.fr/exl-doc/pleins_textes/2022-09/010085334.pdf
%V 36
%W Horizon (IRD)
%X Primary productivity in the Southern Ocean plays a key role in global biogeochemical cycles. While much focus has been placed on phytoplankton production seasonality, non-seasonal fluctuations exceed the amplitude of the seasonal cycle across large swaths of the Antarctic Circumpolar Current. This non-seasonal variability comprises a broad range of timescales from sub-seasonal (<3 months) to multi-annual (>1 year), all of which can project onto the annual mean value. However, year-to-year variations of surface chlorophyll (SChl), a proxy for phytoplankton biomass, are typically attributed to ocean circulation changes associated with the Southern Annular Mode (SAM), which implicitly assumes that sub-seasonal variability averages to near-zero over long timescales. Here, we test this assumption by applying a timeseries decomposition method to satellite-derived SChl in order to separate the low-frequency and high-frequency contributions to the non-seasonal variability. We find that throughout most of the Southern Ocean, year-to-year SChl variations are dominated by the sub-seasonal component, which is not strongly correlated with the SAM. The multi-annual component, while correlated with the SAM, only accounts for about 10% of the total SChl variance. This suggests that changes in annual mean SChl are related to intermittent forcing at small scales, rather than low-frequency climate variability, and thus do not remain correlated over large regions.
%$ 021 ; 036 ; 126